This page contains details about how to create and refine advanced searches when looking for literature in the research databases accessed from the Libraries @ Rochester Regional Health.
For more general information on using databases, please see our page on How to Use Databases.
This page will cover the following topics:
1. Identifying key concepts
2. Controlled vocabulary
3. Field limiters
4. Boolean operators
5. Using quotation marks to search for multiple-word phrases
6. Truncation and wildcards
7. Proximity searching
8. Parentheses
9. Putting it all together
In addition to the content on this page, click the pdf resource below to download a reference sheet that demonstrates the different ways that searches are constructed in individual databases.
This reference sheet will be useful when reviewing the tutorials on this page, and will provide a useful reference afterward, when you make your own searches.
Before starting a search, you have to know what you're searching for! Identify the key concepts you're investigating to keep your search focused and directed.
To identify key concepts, it can be helpful to phrase your subject as an explicit clinical question.
Use a system like PICO to help you identify the core components of the question you're investigating, and use those concepts to inform the keywords you use during a search.
Including subject headings (also called controlled vocabulary) into your search can help you to discover literature on a subject even when it may use different terminology than you're expecting. For example, using the controlled vocabulary term "gastrointestinal hemmorage"[mesh] within PubMed would also bring up articles that call the phenomenon "gastrointestinal bleeding."
MeSH headings are used by MEDLINE databases like PubMed and OVID. A database of MeSH terms can be located at ://www.ncbi.nlm.nih.gov/mesh/, and you can also use this tool to build your searches in PubMed.
Other databases may have their own controlled vocabulary, such as CINAHL subject headings.
Sometimes, searching for a term will yield a great number of irrelevant results because a concept will be mentioned when it is not the focus of an article: for instance, an article could use the word "diabetes" in an incidental manner and appear in your results list when it is not actually focused on that condition.
In such cases, limiting terms to specific fields--for example, specifying "I only want articles that have the word 'diabetes' in their titles"-- can help limit your results to articles that are actually focused on the subject of interest. The 'title' field in particular practically ensures this: few articles are written that mention a specific condition or subject without being majorly focused on it.
The 'abstract' field is more permissive than the 'title' field, while still raising the likelihood of the subject being discussed in some significance.
When you combine field limiters with controlled vocabulary, you can create a search that is comprehensive within a subject of interest while also maintaining a focus on that subject itself. In fact, setting a field limiter is how you search for controlled vocabulary.
Each database has different methods for specifying or limiting fields, and these methods are usually detailed in documentation within a database's help or training pages. For example, PubMed contains a list of fields at https://pubmed.ncbi.nlm.nih.gov/help/#search-tags. Other databases may require you to log in by clicking on their links through the library website in order to find and view their help pages and documentation.
Here are some examples of the different ways to input field limiters between databases:
You may have noticed the use of the word OR above in the example of searching CINAHL for articles with the word "diabetes" in their titles or abstracts. The OR in this example is called a Boolean operator, and such operators provide the foundation for building a complex search.
There are three Boolean operators: AND, OR, and NOT
Placing AND between two words will instruct the database to only look for articles that contain both words.
Placing OR between two words will instruct the database to look for articles that contain either word, even if they do not have both.
Placing NOT between two words will instruct the database to look for articles that contain the first word, but not the second.
For example:
AND
diabetes AND exercise
Results must contain BOTH concepts
OR
diabetes OR exercise
Results can contain EITHER concept
NOT
Diabetes NOT exercise
Results must contain ‘diabetes’ but will be excluded if they contain ‘exercise’
Most databases support the use of Boolean operators. We advise that you enter the words in all capital letters to ensure that the search engine interprets them correctly.
Thus far, we have discussed searches that utilize individual keywords, but often you will need to find phrases that contain multiple words instead of just looking for single-word concepts.
Putting the words gastrointestinal bleeding into a database will cause different interpretations depending on the database (PubMed for instance utilizes algorithms that will attempt to interpret a number of indexed phrases), but most databases will interpret this query as gastrointestinal AND bleeding which would find a great number of articles which contain both words but which are not about the explicit concept of gastrointestinal bleeding.
In cases where you need to search a phrase or concept that is made up of multiple words, you can place those words in quotation marks to search for that exact phrase. Searching for "gastrointestinal bleeding" will yield only articles which contain that exact phrase.
You can also use Boolean operators alongside phrases in quotation marks, for instance "gastrointestinal bleeding" OR "gastrointestinal hemorrhage" will instruct the database to search for articles that contain either phrase; but it will search for those exact phrases and not just the individual words that make them up.
Using quotation marks will search for the exact word or phrase you specify. It will not search for pluralized versions of the word, gerunds, or alternate spellings. Using truncation or wildcards, detailed below, can help with this issue. Otherwise, you will have to iterate each potential variant for the database to search: e.g., "color" OR "colors" OR "coloring" OR "colour" OR "colours" OR "colouring"
Most databases will allow for the use of wildcards and truncation when searching, which can help with iterating on variants of words.
Truncation generally refers to when you place an asterisk at the end of a word to search for variant ways of ending that word. This can be very helpful, but be aware that it can yield results you're not interested in. For instance if you are searching for articles related to nurses and nursing, searching for "nurs*" will yield results that include the words nurse, or nurses, or nursing, but also articles related to nurseries.
Databases often have the ability to search for words within a certain proximity to one another. This can be useful to allow for different ways that authors may word a concept. For instance, a search of "gastrointestinal bleed*" will not find articles that contain a phrase like "bleeding of the gastrointestinal tract."
Each database has very different ways to construct a proximity search. For a useful reference, please see the MEDLINE_Searching_Comparison document linked at the top of this page. More advanced details are contained within the help sections and documentation that can be found within each database.
As an example of how to construct a proximity search in CINAHL, you could use the following string to find articles that contain many potential variants of phrasing the concept of gastrointestinal bleeding: gastrointestinal N3 bleed*.
Of course, this will also potentially find other results that are irrelevant where the two words appear in proximity. For instance, in this example image, the first result is likely irrelevant:
Nevertheless, later in the results set is an article (not pictured) which uses the phrase "gastrointestinal tract bleeding," caught by the proximity search.
Combining all of the techniques and tools detailed in this tutorial, you can experiment and revise your search to find results that balance specificity and broadness and find a result set that contains a tolerable amount of irrelevant results while still remaining comprehensive enough to catch articles of relevance that may be missed by searches that are too strict.
In order to combine several of the functions and strategies we have detailed, you must use parentheses to help databases parse your intentions.
Parentheses in searches work exactly as they do in math problems: each statement contained within parentheses is treated as its own equation and resolved before the elements that exist outside of parentheses. Parentheses can also be nested. For example, ("gastrointestinal hemorrhage"[mesh] OR ("gastro*"[title] AND ("bleed*"[title] OR "hemorrhag*"[title])))
Now it's time to put together what we have covered to create a search.
The general idea that underlies the construction of advanced searches is to identify your key concepts, then think of any synonyms (including controlled vocabulary, proximity variants, etc.) for an individual concept and combine those synonyms with OR.
Place parentheses around the resulting concept-or-synonym group.
Repeat this for the remaining key concepts, and then combine each of these units with the word AND.
Thus: (concept1 OR synonym OR synonym) AND (concept2 OR synonym OR synonym)
Or, for a simple example: (“myocardial infarct*” OR “heart attack*”) AND (“naproxen” OR “aspirin” OR “ibuprofen” OR “NSAID” OR “nonsteroidal anti-inflammatory drug*”)
Combining all these concepts as well as using the built-in filters contained in databases allows you to refine your searches to an incredible degree. It may be impossible to eliminate irrelevant results completely, but using all the tools at your disposal will help you get closer to a search that balances specificity with comprehensiveness and yields relevant literature that can inform your practice.
What follows are two examples of actual searches requested from the library, and the PubMed search strings used to find literature on the subjects requested. The key concepts as identified through PICO are highlighted in different colors to help you see how a concept becomes a search statement within the larger search:
Here are the results, as seen in PubMed:
Here is the second example:
As noted below the example, two of the concepts were connected with OR rather than AND in this example: you should alter your search strategy based on the concepts you're investigating and how they relate to one another!
Here is how that search appears in PubMed:
We hope you found this tutorial on advanced searching techniques helpful. For further assistance, you can find documentation in help files and tutorials within individual databases. For instance, PubMed's User FAQ contains detailed information about how to utilize the database to its full potential:
If you need additional help or want to request a search request from the library, please contact us using the forms available on the right side of this page, or by emailing us directly at library@rochesterregional.org!
Librarians work virtually M-F 8am-4pm.